One useful tool in trying to understand the vast parameter space of variables influencing epilepsy is computational modeling. Computer modeling allows the user to quickly change descriptive variables such as ion channels and synaptic properties, connections, and the temporal characteristics of synaptic messaging. A variety of cell geometries and numbers of neurons have been implemented in an effort to describe various features of the activities of linked cellular networks all in an attempt to reproduce various experimental findings. We propose in this current effort to compare and calibrate the activity derived from a single compartment Hodgkin-Huxley based model with a more sophisticated multicompartmental cellular model where in both cases the cells are arranged and connected in an architecturally realistic fashion to mimic neocortex. Additionally, the simpler single compartment model will be used as the surrounding boundary region for the more precise multicompartment model in an effort to expand the modeled region and still provide meaningful electrophysiologic correlates. We will examine specifically bursting activity derived from a region of the model which acts as an epileptic focus. This comparative work will be used to pursue the following aims: Aim 1) To measure the rate of ictal spread within a calibrated architecturally realistic neuronal network using a hybrid compartment organization, and to perform a sensitivity analysis of this spread as a function of connectivities in the Layer II/III and Layer V pyramidal cell systems. Aim 2) To measure the underlying activity threshold for ictal onset through graded, laminar specific pharmacologic blockade in the context of this model for comparison with experiment. Aim 3) To compare the results obtained in our simulation studies with recordings obtained from a new subdural grid design incorporating standard electrode elements and microwires for local field potential measurements. The goal of these studies is to provide information in a quantitative manner on seizure spatial spread and temporal evolution as a function of neighboring or intrinsic pathology. We plan to model relatively large regions of cortex (6.4 mm X 6.4 mm), while examining the activity of a very high precision smaller (250 <m X 250 <m) representation.